Cues to reduce modulation informational masking

Abstract

The detectability of target amplitude modulation (AM) can be reduced by masker AM in the same carrier-frequency region. It can be reduced even further, however, if the masker-AM rate is uncertain [Conroy and Kidd, J. Acoust. Soc. Am. 149, 3665-3673 (2021)]. This study examined the effectiveness of contextual cues in reducing this latter, uncertainty-related effect (modulation informational masking). Observers were tasked with detecting fixed-rate target sinusoidal amplitude modulation (SAM) in the presence of masker SAM applied simultaneously to the same broadband-noise carrier. A single-interval, two-alternative forced-choice detection procedure was used to measure sensitivity for the target SAM; masker-AM-rate uncertainty was created by randomly selecting the AM rate of the masker SAM on each trial. Relative to an uncued condition, a pretrial cue to the masker SAM significantly improved sensitivity for the target SAM; a cue to the target SAM, however, did not. The delay between the cue-interval offset and trial-interval onset did not affect the size of the masker-cue benefit, suggesting that adaptation of the masker SAM was not responsible. A simple model of within-AM-channel masking captured important trends in the psychophysical data, suggesting that reduced masker-AM-rate uncertainty may have played a relatively minor role in the masker-cue benefit.

FIG. 1.

(A) Group-mean estimates of d′ for each condition at both the low (white bars) and high (black bars) target-modulation depths. (B) Group-mean cue benefits (cued minus uncued d′) for the two cued conditions at both target depths (white bars, low target depth; black bars, high target depth). Error bars in (A) and (B) are standard errors of the mean (SEM). (C) Model d_a's and (D) cue benefits (cued minus uncued d_a) for the different conditions tested in experiment 1 plotted according to the same conventions as the psychophysical results shown in (A) and (B). Model d_a's and cue benefits are the means across four model observers, each of whom completed 1000 trials for each stimulus configuration. Error bars are SEM across the four model observers.

FIG. 2.

Group-mean d′'s (A) and masker-cue benefits (B) for the different conditions tested in experiment 2. For the masker-cue condition, white bars give the group-mean d′ and masker-cue benefit at the 25-ms ISI, whereas black bars give the group-mean d′ and masker-cue benefit at the 500-ms ISI. Otherwise, conventions are the same as in Fig. 1.

FIG. 3.

Group-mean d''s (A) and masker-cue benefits (B) are plotted as a function of the AM rate of the masker SAM on each trial. In (A), black bars give the d''s for the uncued condition, whereas white bars (plotted behind the black bars) give the d''s for the masker-cue condition. The vertical dashed lines in (A) and (B) mark 32 Hz, i.e., the AM rate of the target; hence, the gap between bars in the region of the vertical dashed line demarcates the protected zone. Error bars in (A) and (B) are SEM. (C), (D) Masker-specific estimates of d_a and cue benefits obtained using the single-channel model are plotted according to the same conventions as were used in (A) and (B). The model d_a's and cue benefits represent the group-mean estimates across 11 model observers who completed the same number of trials and same conditions as the psychophysical observers. Error bars in (C) and (D) are SEM.

FIG. 4.

(A) d''s for each of the 11 observers who participated in the study for the uncued (black bars) and masker-cue (white bars) conditions. For each observer, the black bar is plotted on top of the white bar, and so a visible white bar indicates a masker-cue benefit. The observers are rank ordered according to their sensitivity in the uncued condition. (B) Masker-cue benefits are depicted for the 11 observers (same order) appearing in (A).